🌤🌤 CONTRIBUTE 🎉🎉 #50
Labels
contribute
documentation
Improvements or additions to documentation
good first issue
Good for newcomers
🌤🌤目标
首先,任何kernel实现都欢迎,本仓库学习/练习为主,性能最优非本仓库最终目标,先会用,然后再用好。性能最优推荐直接使用cuBLAS, cuDNN, FlashAttention, TensorRT等官方实现。如果有感兴趣的kernel希望在本仓库实现,可以评论本issue(虽然我不一定有能力实现🌚),比如:
☕️☕️Kernel Trace
👨💻👨💻代码规范
提交代码需要遵循以下规范:
🎉🎉 致谢
感谢 @bear-zd, @wangzijian1010等为本仓库提供大量kernel实现 ~
☕️☕️Kernel Trace
The text was updated successfully, but these errors were encountered: